CN112698215A - Estimating battery state from electrical impedance measurements using a convolutional neural network device - Google Patents
Estimating battery state from electrical impedance measurements using a convolutional neural network device Download PDFInfo
- Publication number
- CN112698215A CN112698215A CN202011140740.7A CN202011140740A CN112698215A CN 112698215 A CN112698215 A CN 112698215A CN 202011140740 A CN202011140740 A CN 202011140740A CN 112698215 A CN112698215 A CN 112698215A
- Authority
- CN
- China
- Prior art keywords
- electrical impedance
- series
- neural network
- battery
- convolutional neural
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/367—Software therefor, e.g. for battery testing using modelling or look-up tables
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/382—Arrangements for monitoring battery or accumulator variables, e.g. SoC
- G01R31/3842—Arrangements for monitoring battery or accumulator variables, e.g. SoC combining voltage and current measurements
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/389—Measuring internal impedance, internal conductance or related variables
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/392—Determining battery ageing or deterioration, e.g. state of health
Abstract
The present application relates to estimating battery state from electrical impedance measurements using a convolutional neural network device. A computer-implemented method and battery state estimation system for estimating a battery state of an electrochemical battery, comprising: providing (S12) a series of electrical impedance measurements of the electrochemical cell (14), the series being dependent on a respective measurement frequency (f)s) Sorting; and determining (S20) electrochemistry using the artificial convolutional neural network device (28)A battery state of the battery (14), the artificial convolutional neural network device being configured to receive as input a series of electrical impedance values, wherein the series of electrical impedance values is provided to the artificial convolutional neural network device (28), the series of electrical impedance values corresponding to the provided series of electrical impedance measurements, wherein the artificial convolutional neural network device (28) receives and processes the provided series of electrical impedance values to generate therefrom at least one output signal representative of a battery state (38) associated with the electrochemical battery (14).
Description
Technical Field
The present invention relates to a computer-implemented method of estimating a battery state of an electrochemical cell. Furthermore, the invention relates to a battery state estimation system for estimating a battery state of an electrochemical battery.
For example, the computer may be a microcontroller. For example, a computer or microcontroller may include a processing unit, memory, and input/output ports.
Background
WO 2005/059579 a1 and EP 1702219B 1 describe an apparatus and method for estimating the state of charge of a battery by using a neural network. The apparatus includes a sensing part for detecting current, voltage and temperature from the battery cell, and a neural network performing a neural network algorithm and a learning algorithm based on data of the current, voltage and temperature transmitted thereto from the sensing part and current time data. In one example, a dynamic multi-dimensional wavelet neural network includes an input domain, a hidden layer, and an output layer.
US 4678998A describes a battery condition monitor and monitoring method.
JP 2003-249271A and JP 4038788B 2 describe the determination of the state of degradation and residual capacity of a battery in real time. In one example, the measurement unit may periodically measure and sample the voltage, current, internal impedance, and temperature of the battery as the operating parameters of the battery in operation. The internal impedance of the battery was measured by adding an AC signal of 1kHz and 100mA to the battery. The first neural network may determine the degradation state as "normal", "reminder (accounting)" and "degradation" based on the operation parameter of the battery from the measurement unit, and the second neural network may determine the residual capacity based on the operation parameter of the battery and the degradation state from the first neural network. Each of the first and second neural networks is a three-layer feed-forward network having an input layer, a middle layer, and an output layer.
US 6307378B 1 describes a method and apparatus for measuring the impedance of an electrochemical cell (cell) and battery (battery).
WO 03/071617 a2 describes a method for determining a condition parameter of an electrochemical cell, such as a battery. In one example, measurement signals such as terminal voltage, battery voltage, load current, charge current, ambient temperature, battery surface temperature, terminal temperature, internal battery temperature, and impedance signals are passed to a feature extraction processing algorithm that generates feature vectors and feature signatures. Data from the feature vectors are passed to a neural network ISOC predictor for initial battery capacity state of charge estimation, and a neural network CSOC predictor for continuous prediction of SOC during operation. In another example, the information contained in the feature vectors is used for state of health classification by neural network SOH classifiers, linear/statistical SOH classifiers, and fuzzy logic SOH classifiers. Neural networks designed for direct SOC estimation use a hidden layer.
WO 2016/208745 a1 and its translation DE 112016002873T 5 describe a method of identifying the state of charge or the depth of discharge of a battery. The method includes determining a complex impedance between a positive pole and a negative pole of the battery for a plurality of frequencies.
WO 2017/110437 a1 and its translation DE 112016003789T 5 describe an estimation device for estimating the residual capacity of a lithium ion battery.
WO 2016/080111 a1 and its translation DE 112015005201T 5 describe an estimation device for estimating the remaining storage capacity of a battery.
US 2013/0307487 a1 and US 8994340B 2 describe a method and system for determining the temperature of cells in a battery pack by measuring the impedance of the cells (cells) and using the impedance to determine the temperature without the use of temperature sensors.
Summary of The Invention
It is an object of the present invention to provide a new method of estimating the cell state of an electrochemical cell based on electrical impedance measurements. It is desirable that the method better utilizes the information comprised in the series of electrical impedance measurements for different measurement frequencies.
The invention is specified in the independent claims. Further embodiments are specified in the dependent claims.
According to one aspect of the invention, there is provided a computer-implemented method of estimating a battery state of an electrochemical cell, the method comprising:
providing a series of electrical impedance measurements of the electrochemical cell, each electrical impedance measurement being measured at a respective measurement frequency, the series being ordered according to the respective measurement frequency,
determining a battery state of an electrochemical battery using an artificial convolutional neural network device, the artificial convolutional neural network device configured to receive as an input a series of electrical impedance values,
wherein a series of electrical impedance values is provided to the artificial convolutional neural network device, the series of electrical impedance values corresponding to the series of electrical impedance measurements provided,
wherein the artificial convolutional neural network device receives and processes the provided series of electrical impedance measurements to generate therefrom at least one output signal indicative of a battery state associated with the electrochemical cell.
The series of electrical impedance values corresponds to the series of electrical impedance measurements provided. That is, the two series characterize the same curve progression of the electrical impedance in the complex plane with respect to the indices of the elements of the respective series. In this context, the term "complex plane" should be understood as a complex plane of electrical impedance, unless otherwise specified.
In particular, a series of electrical impedance values is provided based on the provided series of electrical impedance measurements.
For example, the provided series of electrical impedance values may be the same as or included in or generated from the provided series of electrical impedance measurements.
For example, generating a series of electrical impedance values from the provided series of electrical impedance measurements may include at least one of: interpolating between the electrical impedance measurements, selecting from the electrical impedance measurements, and extrapolating the electrical impedance measurements.
The artificial convolutional neural network device may receive and process the series of electrical impedance values to generate an output signal therefrom according to a predetermined processing structure of the artificial convolutional neural network device.
For example, the artificial convolutional neural network device may include a Convolutional Neural Network (CNN).
The CNN has an input layer, at least one convolutional layer, and an output layer. The CNN may be a deep neural network. The CNN may be trained to recognize patterns in the transformed series of electrical impedance measurements and associate the patterns with corresponding battery states.
For example, the artificial convolutional neural network device may have been trained to estimate the battery state of a (rechargeable) electrochemical battery by detecting characteristic characteristics of a series of electrical impedance values, using temperatures for a predetermined temperature range and training data for different states of health and/or different states of charge of the battery.
Convolutional neural networks are particularly well suited to recognizing patterns that appear in input vectors or input arrays due to convolutions performed within the convolutional layers of the convolutional neural networks. For example, convolutional neural networks are known for classifying images.
It has been found that convolutional neural networks, due to their pattern recognition capabilities, are suitable for estimating the state of a battery associated with an electrochemical cell based on a series of electrical impedance values corresponding to a series of electrical impedance measurements, the series being ordered according to respective measurement frequencies.
In contrast, in the prior art of WO 03/071617 a2, the neural network designed for direct SOC estimation uses one hidden layer, and data from several different measurement signals is included in the feature vectors, or passed to a feature extraction processing algorithm that generates the feature vectors; the data includes electrical parameters such as battery voltage and current, and temperature. Thus, WO 03/071617 a2 does not disclose the application of convolution to such data of feature vectors.
Preferably, the electrochemical cell is a rechargeable electrochemical cell.
Estimating the battery state in the form of a state of charge is very valuable, for example, for mobile appliances such as mobile tools or electrically driven mobile devices such as vehicles.
Similarly, estimation of battery state in the form of state of health is very important for the reliability of the device.
For many applications, in particular rechargeable electrochemical cells, the knowledge of the cell state in the form of the cell temperature of the electrochemical cell is also of high importance.
By estimating the battery state of the battery using an artificial convolutional neural network device, the battery state may be estimated based on a series of provided electrical impedance measurements (i.e., based on a corresponding series of electrical impedance values). In particular, the battery state may be estimated based only on values corresponding to direct measurements, such as direct electrical measurements (such as a series of only electrical impedance measurements provided).
The electrical impedance may also be referred to as complex electrical impedance and may be complex and may be expressed in units of resistance (e.g., ohms).
For example, each electrical impedance measurement in the series of electrical impedance measurements provided may be or include a complex number defining real and imaginary parts of the respective electrical impedance. The electrical impedance may also be expressed in polar form, defining an amplitude (or amplitude) and an angle (or phase).
For example, the electrical impedance measurements in the series of electrical impedance measurements provided may be in the form of a corresponding complex representation (complex number). Alternatively, for example, the electrical impedance measurements may each be in the form of amplitude and phase, and the providing step may comprise converting the electrical impedance measurements into electrical impedance measurements in the form of respective complex representations of the electrical impedance measurements. The "complex representation" of the electrical impedance measurement includes a complex number. For example, the complex representation of the electrical impedance measurement may be a complex number.
Herein, the term "battery" should be understood to include battery cells. A battery may include one or more battery cells. In particular, the term "battery" includes a battery cell as well as a battery composed of a plurality of battery cells.
In particular, an electrochemical cell is understood to define a device consisting of one or more electrochemical cells with external electrical connections. For example, the battery may comprise two external electrical connections for drawing power from one or more electrochemical cells and, in the case of a rechargeable electrochemical cell, for (re) charging one or more electrochemical cells.
For example, the corresponding measurement frequency may be the frequency of the signal input to the electrochemical cell. For example, the signal may be a sinusoidal signal.
Preferably, each electrical impedance value in the provided series of electrical impedance values comprises a complex number defining a real part and an imaginary part of the respective electrical impedance.
Elements of the series of electrical impedance values are fed to corresponding inputs of the artificial convolutional neural network device. Thus, automatic battery state estimation based on the series of electrical impedance values provided is possible.
Preferably, the series of electrical impedance measurements is provided in the form of digital signals.
Preferably, the series of electrical impedance measurements of the electrochemical cell is provided in the form of a digital representation.
For example, the series of electrical impedance measurements may be received from an electrical impedance measurement unit or an electrical impedance measurement device.
Providing the series of electrical impedance measurements may include receiving the electrical impedance measurements in the series one after another.
The series of electrical impedance measurements may be transmitted to a battery state estimation system comprising means for performing the method steps. The system may be configured to provide (including receive) the transmitted electrical impedance measurements.
In one or more embodiments, the measurement frequencies are assumed to be logarithmically equally spaced. Preferably, the measurement frequencies are equally logarithmically spaced over at least four decimal places (decades) of the measurement frequency range (frequency range of the measurement frequencies), and more preferably over at least five decimal places of the measurement frequency range.
Preferably, the measurement frequency comprises a measurement frequency in the range of 0.1 to 1.0 Hz.
Preferably, the measurement frequency comprises a measurement frequency in the range of 1kHz to 10 kHz.
Preferably, the measurement frequencies comprise at least 4 (four) measurement frequencies per decade of the measurement frequency range, more preferably at least five measurement frequencies per decade of the measurement frequency range.
The step of providing a series of electrical impedance measurements may comprise receiving a plurality of electrical impedance measurements, each electrical impedance measurement being measured at a respective measurement frequency, and:
-ranking the plurality of electrical impedance measurements according to the associated measurement frequency to provide a series of electrical impedance measurements, or
-providing a plurality of electrical impedance measurements as a series of electrical impedance measurements.
For example, a ranked plurality of electrical impedance measurements optionally including respective measurement frequencies may be provided as a series of electrical impedance measurements.
A plurality of electrical impedance measurements or series of electrical impedance measurements may be received in the form of an electrical impedance spectrum.
For example, in a series of electrical impedance measurements, the electrical impedance measurements may include a corresponding measurement frequency. For example, each electrical impedance measurement may include a measurement value of electrical impedance and a measurement frequency. However, the electrical impedance measurements may also be in the form of corresponding electrical impedances, expressed as corresponding complex numbers, or both expressed as amplitudes and phases.
An output signal indicative of a battery state of the electrochemical cell is generated. For example, the at least one output signal may represent a classification and/or estimation of a battery state.
The battery status may include at least one of: state of charge (SoC) of the electrochemical cell, state of health (SoH) of the electrochemical cell, state of function (SoF) of the electrochemical cell, capacity of the electrochemical cell, and temperature of the electrochemical cell.
For example, the output signal may be transmitted or conveyed or output to an output unit for outputting the signal, and/or for outputting a visual signal based on the output signal. The visual signal may be displayed.
Preferably, the series of electrical impedance values provided has a predetermined number of elements. For example, the number of elements may correspond to the number of inputs of an artificial convolutional neural network device for receiving a series of electrical impedance values.
For example, generating the series of electrical impedance values from the provided series of electrical impedance measurements may include adjusting a number of elements of the series of electrical impedance measurements to a predetermined number of elements.
For example, a series of electrical impedance values may be generated from a provided series of electrical impedance measurements by adjusting the number of elements of the series of electrical impedance measurements to a predetermined number of elements.
For example, the method may include: if the number of elements of the series of electrical impedance measurements provided is different from the predetermined number of elements, the number of elements of the series of electrical impedance measurements is adjusted to the predetermined number of elements.
In one or more embodiments, the method may include adjusting the number of elements of the series of electrical impedance measurements or adjusting the number of elements of the series of electrical impedance values to a predetermined number of elements.
Thus, the number of elements may be adjusted to the number of corresponding inputs of the artificial convolutional neural network device for receiving the series of electrical impedance values.
For example, the predetermined number of elements may be a predetermined number of elements of a series of electrical impedance values.
For example, the number of elements of the series of electrical impedance values may be adjusted to a predetermined number of elements by adjusting the number of elements of the series of electrical impedance measurements to the predetermined number of elements.
For example, adjusting the number of elements of the series of electrical impedance measurements or the number of elements of the series of electrical impedance values may comprise at least one of: interpolating between the electrical impedance measurements, selecting from the electrical impedance measurements, and extrapolating the electrical impedance measurements. In the adjusting step, the number of elements may be increased, maintained, or decreased.
In one or more embodiments, the artificial convolutional neural network device receives and processes the series of electrical impedance values to generate therefrom at least one output signal representative of the state of charge of the electrochemical cell.
In one or more embodiments, the artificial convolutional neural network device receives and processes the series of electrical impedance values to generate therefrom at least one output signal representative of the state of health of the electrochemical cell.
In one or more embodiments, the artificial convolutional neural network device receives and processes the series of electrical impedance values to generate therefrom at least one output signal representative of the functional state of the electrochemical cell.
In one or more embodiments, the artificial convolutional neural network device receives and processes the series of electrical impedance values to generate therefrom at least one output signal representative of the temperature of the electrochemical cell.
In one or more embodiments, the method further comprises:
calculating an electrical impedance gradient from the series of electrical impedance measurements to generate a series of electrical impedance gradients,
wherein the artificial convolutional neural network device is configured to receive the series of electrical impedance gradients as a further input,
wherein the artificial convolutional neural network device receives and processes at least the provided series of electrical impedance measurements and the series of electrical impedance gradients to generate therefrom at least one output signal indicative of a battery state associated with the electrochemical cell.
Thus, a series of electrical impedance gradients can be provided for evaluation by the artificial convolutional neural network device, which emphasizes information about the variation of electrical impedance as a function of the measurement frequency.
For example, the electrical impedance gradient of a series of electrical impedance measurements may be calculated with respect to the measurement frequency, with respect to the logarithm of the measurement frequency, or with respect to the index of the elements of the series of measurements. That is, the respective gradients at the respective indices of the elements of the series of measurements may be calculated as a change in electrical impedance as a function of a change in measurement frequency, as a function of a logarithm of a measurement frequency, or as a function of a change in index of an element of the series of measurements.
For example, for a respective element of the series of electrical impedance measurements, a corresponding gradient may be calculated to express a change in the value of the electrical impedance measurement at the respective measurement frequency, its logarithm, or the respective element index as a function of a change in the measurement frequency, as a function of a change in the logarithm of the measurement frequency, or as a function of a change in the element index. Thus, the gradient may be calculated for a series of discrete electrical impedance measurements. Calculating the gradient may correspond to differentiating the continuous electrical impedance curve with respect to the measurement frequency or with respect to the logarithm of the measurement frequency or the index of the elements of the series of measurements. For example, for a respective element of the series of electrical impedance measurements, a corresponding gradient may be calculated based on the difference between the element and an adjacent element in the series of electrical impedance measurements divided by the difference of the respective index of the element or the respective measurement frequency.
For example, calculating the gradient of the series of electrical impedance measurements may comprise calculating the gradient of the real part and the gradient of the imaginary part of the series of electrical impedance measurements.
In one or more embodiments, the method comprises:
the electrical impedance of the electrochemical cell is measured at different measurement frequencies using an electrical impedance measurement device to provide a series of electrical impedance measurements.
For example, a series of electrical impedance measurements may be provided based on the electrical impedance measured at different measurement frequencies.
The electrical impedance measurements may be measured in any order and/or simultaneously for the respective measurement frequencies.
For example, the electrical impedance measurement device may be configured to input a signal comprising a respective frequency to the electrochemical cell and determine a ratio of the amplitude and phase of the response signal to the amplitude and phase of the input signal at the same frequency, the frequency of the signal and the frequency of the response signal corresponding to the measurement frequency.
For example, the electrical impedance measurement device may be an electrical impedance spectroscopy device configured to measure the electrical impedance of the electrochemical cell at a range of measurement frequencies. That is, the electrical impedance of the electrochemical cell is measured according to the process of electrochemical impedance spectroscopy.
Using electrochemical impedance spectroscopy, the electrochemical process in an electrochemical cell is characterized by electrical measurements that characterize the AC response of the electrochemical cell to an applied AC signal. In addition to the structural configuration and the connector configuration, chemical processes in the battery can also lead to a characteristic frequency dependence of the measured impedance.
During electrical impedance measurement at a particular measurement frequency, a DC offset signal (offset voltage or offset current) or DC bias signal may be applied to the battery, which is modulated by an AC signal at the measurement frequency.
For example, the series of electrical impedance measurements may be provided from a separate measurement circuit or electrical impedance measurement device. However, the battery state estimation system may also include an electrical impedance measurement device for measuring and providing a series of electrical impedance measurements from the electrochemical cell.
For example, the electrochemical cell may be a lithium ion cell or a lead acid cell.
According to one aspect of the invention, a computer-implemented method of monitoring a battery state of an electrochemical cell may be provided. The monitoring method may comprise the steps of a method of estimating the state of a battery.
For example, the method may be implemented in a battery monitoring system for monitoring a battery condition of an electrochemical cell.
For example, the method may be implemented in a battery charging system for recharging a rechargeable electrochemical cell.
According to one aspect of the present invention, there is provided a battery state estimation system for estimating a battery state of an electrochemical battery, the system comprising means for performing the steps of the method.
For example, the system may be or may be included in a system for monitoring the battery status of an electrochemical cell.
For example, the system may be or may be included in a battery charging system for recharging a rechargeable electrochemical battery.
In one or more embodiments, the battery state estimation system further includes: an electrical impedance measurement device configured to measure the electrical impedance of the electrochemical cell at different measurement frequencies to provide a series of electrical impedance measurements.
Drawings
Preferred embodiments of the present invention will now be described with reference to the accompanying drawings, in which:
FIG. 1 is a schematic diagram of a method of estimating a battery state of a rechargeable electrochemical cell;
FIG. 2 is a schematic graph showing a series of electrical impedance measurements of a rechargeable electrochemical cell measured at respective measurement frequencies and at respective cell temperatures;
FIG. 3 is a schematic diagram showing a series of electrical impedance measurements of a rechargeable electrochemical cell measured at respective measurement frequencies and at respective states of health of the cell;
FIG. 4 is a schematic diagram of determining a gradient of electrical impedance; and
fig. 5 is a schematic diagram of a system for estimating a battery state of a rechargeable electrochemical cell.
Detailed Description
Fig. 1 schematically illustrates a computer-implemented method of estimating a battery state of a rechargeable electrochemical battery (e.g., a lithium-ion battery). For example, the method may be performed by a battery state estimation system as further described below with reference to fig. 5.
Step S10 is a step of measuring the electrical impedance of the electrochemical cell at different measurement frequencies using an electrical impedance measurement device.
In step S12, by measuring the electrical impedance, a series of electrical impedance measurements of the electrochemical cell is provided, for example as a data set, in the form of digital signals. The series is ordered according to the respective measurement frequency, preferably according to the order of increasing measurement frequencies.
However, the method may also start with step S12 providing a measurement result, which may be measured independently of the method and may be transmitted to a computer performing the method.
In case the provided electrical impedance measurements are not already in the form of complex numbers (representing complex impedances), the method may comprise an optional step S14 of converting the provided electrical impedance measurements into complex numbers.
In optional step S16, the number of elements of the series of electrical impedance measurements is adjusted to a predetermined number of elements, for example, to a number of 21 elements.
In step S18, the series of electrical impedance measurements is provided as a series of electrical impedance values to an artificial convolutional neural network device configured to receive as input the series of electrical impedance values.
In step S20, the artificial convolutional neural network device processes the series of electrical impedance values to generate therefrom an output signal indicative of the state of the battery. Thus, based on the series of electrical impedance values, a cell state of the electrochemical cell is determined. In step S22, the battery state is output. For example, the battery state may be a state of health of the battery.
Fig. 2 exemplarily shows four series of electrical impedance measurements measured at respective temperatures T of the battery. Each series being included at a respective measuring frequency fsMeasured electrical impedance measurements. The measurement results of each series are represented by circles with corresponding patterns.
Fig. 2 is a nyquist plot in the form of a two-dimensional graph of imaginary part im (Z) and real part re (Z) of electrical impedance Z. For illustrative purposes, elements of a series are connected by a line. According to convention, the imaginary part is shown in the opposite direction, with the imaginary part increasing towards the bottom of fig. 2. In FIG. 2, arrow fsThe measurement results 25 are shown exemplarily with increasing measurement frequency fsThe order of (a).
Preferably, the measurements are taken at a measurement frequency of logarithmic progression. Preferably, the series of measurements comprises at least 4 (four) measurements per decade of the measurement frequency range. In fig. 2, the electrical impedance measurements of the series of electrical impedance measurements are shown schematically for illustrative purposes only. The number of measurements shown in fig. 2 may deviate from the number of measurements actually used and is for illustrative purposes only.
As shown in fig. 2, the series of curves of the electrical impedance measurements in the complex plane show a large variation for different temperatures T. Typically, this variation is highly non-linear with respect to temperature T.
The multiple series of measurements shown in fig. 2 correspond to different temperatures, but to the same or similar state of health and the same or similar state of charge of the battery. However, as shown in fig. 3, different series a to G of electrical impedance measurements are provided for different states of health (SoH) at the same or similar temperature and the same or similar state of charge. Fig. 3 schematically shows a series of electrical impedance measurements measured at different states of health of the battery. These series are schematically shown by the connecting lines.
Therefore, the curve progression of the electrical impedance measurements varies according to the temperature and the state of health of the battery.
In addition to the series of electrical impedance values, a series of electrical impedance gradients may be provided as a further input to the artificial convolutional neural network device.
Fig. 4 schematically shows a part of a series of electrical impedance measurements connected by a line. The series being included at a measuring frequency fsFirst measurement result obtained at the following, and at the measurement frequency fs+1The second measurement taken next, where s and s +1 represent the corresponding indices in the series of measurements. The real and imaginary parts of the difference between the measurements are denoted as Δ Rs、△Is。
For the corresponding measuring frequency fsThe electrical impedance gradient is calculated with respect to the measurement frequency as follows: the real part of the electrical impedance gradient is calculated as follows: delta Rs/(fs+1-fs) (ii) a The imaginary part of the electrical impedance gradient is calculated as follows: delta Is/(fs+1-fs)。
Thus, the gradient of the series of electrical impedance measurements with respect to the measurement frequency is calculated.
In another embodiment, the gradient may be calculated relative to the elemental index s in the series of measurements as follows: the real part of the electrical impedance gradient can be calculated as: delta Rs/((s+1)-s)＝△Rs(ii) a The imaginary part of the electrical impedance gradient can be calculated as: delta Is/((s+1)-s)＝△Is。
In yet another embodiment, the gradient can be calculated logarithmically with respect to the measured frequency as follows: the real part of the electrical impedance gradient is calculated as follows: delta Rs/(logB(fs+1)-logB(fs) ); the imaginary part of the electrical impedance gradient is calculated as follows: delta Is/(logB(fs+1)-logB(fs) ); log thereinBLogarithm to base B; for example, B ═ 10.
Fig. 5 schematically shows an example of a battery state estimation system 10 configured for performing the method of fig. 1, optionally including an electrical impedance measurement device 12. For example, the battery state estimation system 10 may be implemented in a computer (such as a microcontroller). For example, a microcontroller including the system 10 and optional electrical impedance measurement device 12 may be part of a battery monitoring system for monitoring the battery status of the electrochemical cell 14.
The electrical impedance measurement device 12 includes an electrical impedance measurement unit 16. The battery state estimation system 10 also includes a preprocessing unit 20 and a computing device 22.
For a series of measurement frequencies fsThe electrical impedance measuring unit 12 applies an excitation signal, for example a corresponding measuring frequency f, to the electrochemical cell 14 to be measuredsOf the sinusoidal signal. The signal is an input in the form of a small amplitude Alternating Current (AC) signal and the AC response from the battery 14 is measured. For example, a current signal is input and a voltage response signal is measured. Alternatively, a voltage signal is input and a current response signal is measured. During the measurement, a Direct Current (DC) bias voltage or a DC bias current may be applied depending on the type of electrochemical cell 14. As is known, the measurement settings correspond to Electrochemical Impedance Spectroscopy (EIS) measurement settings. According to the predetermined measurement settings for the electrochemical cell 14, for the respective measurementAs a result, the measurement frequencies are arranged or increased in equidistant steps on a logarithmic scale.
The electrical impedance measured at a particular measurement frequency is the ratio of the amplitude and phase of the AC response signal to the amplitude and phase of the input signal and is represented as a complex number (complex impedance). For example, four different measurement frequencies may be used per decade of measurement frequency.
The pre-processing unit 20 comprises a normalising means 24 for providing a series of electrical impedance measurements from the electrical impedance measurement device 12 and for adjusting the number of elements of the series of electrical impedance measurements to a predetermined number of elements, for example 21 elements. For example, the number of elements may be adjusted by interpolating the series of elements. In the event that the series of electrical impedance measurements provided by the normalizing means 24 already has a target value for a predetermined number of elements, the normalizing means 24 maintains the number of elements.
The pre-processing unit 20 further comprises a gradient calculation means 26 which receives the normalized series of electrical impedance measurements from the normalization means 24. The gradient calculation means 26 calculates the gradient of the series of electrical impedance measurements with respect to the measurement frequency to generate a series of electrical impedance gradients similar to that explained above with reference to figure 4. Thus, the gradient calculation means 26 generates a series of electrical impedance gradients from the normalised series of electrical impedance measurements.
The computing device 22 includes an artificial convolutional neural network device 28 having a first input device 30 for receiving the normalized series of electrical impedance measurements from the preprocessing unit 20 as a series of electrical impedance values.
Furthermore, the artificial convolutional neural network device 28 has a second input device 32 for receiving the series of electrical impedance gradients from the gradient calculation device 26.
For example, the series of electrical impedance values and the series of electrical impedance gradients together may form an input (such as an input vector or an input array) to the artificial neural network device 28.
In addition, the artificial convolutional neural network device 28 includes an output device 36 for outputting an output signal 38 indicative of a battery state (e.g., state of health) associated with the electrochemical cell 14. The artificial convolutional neural network device 28 receives and processes the series of electrical impedance values and the series of electrical impedance gradients and generates an output signal therefrom.
The artificial convolutional neural network device 28 has been trained to estimate the battery state 38 of the electrochemical cell 14 by detecting characteristic features of a series of electrical impedance values and a series of electrical impedance gradients. The determined battery status 38 is output by the output device 36.
The system may also be implemented with an artificial convolutional neural network device 28 having only a first input device 30 for receiving a standardized series of electrical impedance measurements. In this case, the preprocessing unit 20 preferably does not comprise the gradient calculation means 26.
Claims (10)
1. A computer-implemented method of estimating a battery state of an electrochemical cell, the method comprising:
providing (S12) a series of electrical impedance measurements of the electrochemical cell (14), each electrical impedance measurement being at a respective measurement frequency (f)s) Is measured, said series being dependent on said corresponding measuring frequency (f)s) The order is given to the user,
determining (S20) a battery state of the electrochemical battery (14) using an artificial convolutional neural network device (28) configured to receive as input a series of electrical impedance values,
wherein the series of electrical impedance values is provided to the artificial convolutional neural network device (28), the series of electrical impedance values corresponding to the provided series of electrical impedance measurements,
wherein the artificial convolutional neural network device (28) receives and processes the provided series of electrical impedance values to generate therefrom at least one output signal representative of a battery state (38) associated with the electrochemical cell (14).
2. The method of claim 1, wherein the method further comprises:
adjusting (S16) the number of elements of the series of electrical impedance measurements to a predetermined number of elements.
3. The method according to claim 1 or 2, wherein the artificial convolutional neural network device (28) receives and processes the series of electrical impedance values to generate therefrom at least one output signal representative of the state of charge of the electrochemical cell (14).
4. The method according to any one of the preceding claims, wherein the artificial convolutional neural network device (28) receives and processes the series of electrical impedance values to generate therefrom at least one output signal representative of the state of health of the electrochemical cell (14).
5. The method according to any one of the preceding claims, wherein the artificial convolutional neural network device (28) receives and processes the series of electrical impedance values to generate therefrom at least one output signal representative of the functional status of the electrochemical cell (14).
6. The method according to any one of the preceding claims, wherein the artificial convolutional neural network device (28) receives and processes the series of electrical impedance values to generate therefrom at least one output signal representative of a temperature associated with the electrochemical cell (14).
7. The method according to any one of the preceding claims, wherein the method further comprises:
calculating an electrical impedance gradient from the series of electrical impedance measurements to generate a series of electrical impedance gradients,
wherein the artificial convolutional neural network device (28) is configured to receive the series of electrical impedance gradients as a further input,
wherein the artificial convolutional neural network device (28) receives and processes at least the provided series of electrical impedance measurements and the series of electrical impedance gradients to generate therefrom at least one output signal representative of a battery state (38) associated with the electrochemical cell (14).
8. The method according to any one of the preceding claims, wherein the method further comprises:
at different measuring frequencies (f)s) Measuring (S10) the electrical impedance of the electrochemical cell (14) to provide a series of electrical impedance measurements.
9. A battery state estimation system for estimating a battery state of an electrochemical battery (14), the system comprising means for performing the steps of the method according to any one of claims 1 to 8.
10. The battery state estimation system of claim 8, further comprising:
an electrical impedance measurement device (12) configured for measuring at different measurement frequencies (f)s) The electrical impedance of the electrochemical cell (14) is measured to provide a series of said electrical impedance measurements.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| EP19204756.1 | 2019-10-23 | ||
| EP19204756.1A EP3812783A1 (en) | 2019-10-23 | 2019-10-23 | Estimating a battery state from electrical impedance measurements using convolutional neural network means |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112698215A true CN112698215A (en) | 2021-04-23 |
Family
ID=68342592
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011140740.7A Pending CN112698215A (en) | 2019-10-23 | 2020-10-22 | Estimating battery state from electrical impedance measurements using a convolutional neural network device |
Country Status (3)
| Country | Link |
|---|---|
| US (1) | US20210123979A1 (en) |
| EP (1) | EP3812783A1 (en) |
| CN (1) | CN112698215A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112180210A (en) * | 2020-09-24 | 2021-01-05 | 华中科技大学 | Power distribution network single-phase earth fault line selection method and system |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114636933B (en) * | 2022-05-23 | 2022-09-13 | 长沙矿冶研究院有限责任公司 | Composite neural network-based retired power battery capacity detection method and system |
| EP4296701A1 (en) | 2022-06-24 | 2023-12-27 | Novum engineerING GmbH | Battery test system, battery test bench and server and method for assessing a battery state |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101131417A (en) * | 2006-08-22 | 2008-02-27 | 德尔菲技术公司 | Battery monitoring system |
| JP2013519893A (en) * | 2010-02-17 | 2013-05-30 | イエフペ エネルジ ヌヴェル | In-situ battery diagnostic method by electrochemical impedance spectroscopy |
| CN106289566A (en) * | 2016-07-19 | 2017-01-04 | 清华大学 | A kind of method secondary cell internal temperature estimated based on electrochemical impedance |
| US20180067169A1 (en) * | 2016-09-06 | 2018-03-08 | Primearth Ev Energy Co., Ltd. | Battery capacity measuring device and battery capacity measuring method |
| US20180143257A1 (en) * | 2016-11-21 | 2018-05-24 | Battelle Energy Alliance, Llc | Systems and methods for estimation and prediction of battery health and performance |
| CN109143108A (en) * | 2018-07-25 | 2019-01-04 | 合肥工业大学 | A kind of estimation method of the lithium ion battery SOH based on electrochemical impedance spectroscopy |
Family Cites Families (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPS61170678A (en) | 1985-01-25 | 1986-08-01 | Nissan Motor Co Ltd | Battery state detector |
| US6037777A (en) | 1998-09-11 | 2000-03-14 | Champlin; Keith S. | Method and apparatus for determining battery properties from complex impedance/admittance |
| US6307378B1 (en) | 2000-01-03 | 2001-10-23 | The Penn State Research Foundation | Method and apparatus for measurement of electrochemical cell and battery impedances |
| US20030184307A1 (en) | 2002-02-19 | 2003-10-02 | Kozlowski James D. | Model-based predictive diagnostic tool for primary and secondary batteries |
| JP4038788B2 (en) | 2002-02-22 | 2008-01-30 | アクソンデータマシン株式会社 | Battery remaining capacity determination method and apparatus |
| CA2550072C (en) | 2003-12-18 | 2011-04-19 | Lg Chem, Ltd. | Apparatus and method for estimating state of charge of battery using neural network |
| US8994340B2 (en) | 2012-05-15 | 2015-03-31 | GM Global Technology Operations LLC | Cell temperature and degradation measurement in lithium ion battery systems using cell voltage and pack current measurement and the relation of cell impedance to temperature based on signal given by the power inverter |
| US9536293B2 (en) * | 2014-07-30 | 2017-01-03 | Adobe Systems Incorporated | Image assessment using deep convolutional neural networks |
| JP6555773B2 (en) | 2014-11-18 | 2019-08-07 | 学校法人立命館 | Storage power remaining amount estimation device, method for estimating remaining power storage amount of storage battery, and computer program |
| US10534038B2 (en) | 2015-06-26 | 2020-01-14 | Japan Aerospace Exploration Agency | Method and system for estimating state of charge or depth of discharge of battery, and method and system for evaluating health of battery |
| JP6323441B2 (en) | 2015-12-25 | 2018-05-16 | マツダ株式会社 | Lithium-ion battery remaining capacity estimation device |
| WO2019021099A1 (en) * | 2017-07-25 | 2019-01-31 | 株式会社半導体エネルギー研究所 | Power storage system, electronic apparatus, vehicle, and estimation method |
| WO2019021095A1 (en) * | 2017-07-26 | 2019-01-31 | 株式会社半導体エネルギー研究所 | System for controlling charging of secondary cell and method for detecting abnormality in secondary cell |
| US20210033675A1 (en) * | 2018-03-20 | 2021-02-04 | Gs Yuasa International Ltd. | Degradation estimation apparatus, computer program, and degradation estimation method |
| JP7157908B2 (en) * | 2018-12-20 | 2022-10-21 | トヨタ自動車株式会社 | Battery capacity estimation method and battery capacity estimation device |
| US11574223B2 (en) * | 2019-10-07 | 2023-02-07 | Intelligent Fusion Technology, Inc. | Method and apparatus for rapid discovery of satellite behavior |
| EP4046061A1 (en) * | 2019-10-14 | 2022-08-24 | Ventana Medical Systems, Inc. | Weakly supervised multi-task learning for cell detection and segmentation |
-
2019
- 2019-10-23 EP EP19204756.1A patent/EP3812783A1/en active Pending
-
2020
- 2020-10-22 CN CN202011140740.7A patent/CN112698215A/en active Pending
- 2020-10-22 US US17/077,919 patent/US20210123979A1/en not_active Abandoned
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101131417A (en) * | 2006-08-22 | 2008-02-27 | 德尔菲技术公司 | Battery monitoring system |
| JP2013519893A (en) * | 2010-02-17 | 2013-05-30 | イエフペ エネルジ ヌヴェル | In-situ battery diagnostic method by electrochemical impedance spectroscopy |
| CN106289566A (en) * | 2016-07-19 | 2017-01-04 | 清华大学 | A kind of method secondary cell internal temperature estimated based on electrochemical impedance |
| US20180067169A1 (en) * | 2016-09-06 | 2018-03-08 | Primearth Ev Energy Co., Ltd. | Battery capacity measuring device and battery capacity measuring method |
| US20180143257A1 (en) * | 2016-11-21 | 2018-05-24 | Battelle Energy Alliance, Llc | Systems and methods for estimation and prediction of battery health and performance |
| CN109143108A (en) * | 2018-07-25 | 2019-01-04 | 合肥工业大学 | A kind of estimation method of the lithium ion battery SOH based on electrochemical impedance spectroscopy |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112180210A (en) * | 2020-09-24 | 2021-01-05 | 华中科技大学 | Power distribution network single-phase earth fault line selection method and system |
| CN112180210B (en) * | 2020-09-24 | 2021-08-10 | 华中科技大学 | Power distribution network single-phase earth fault line selection method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| US20210123979A1 (en) | 2021-04-29 |
| EP3812783A1 (en) | 2021-04-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112698215A (en) | Estimating battery state from electrical impedance measurements using a convolutional neural network device | |
| US20200081070A1 (en) | Neural-network state-of-charge and state of health estimation | |
| Wu et al. | A novel state of health estimation method of Li-ion battery using group method of data handling | |
| US11598816B2 (en) | Method and system for online assessing state of health of a battery | |
| JP7157909B2 (en) | Battery capacity estimation method and battery capacity estimation device | |
| JP7157908B2 (en) | Battery capacity estimation method and battery capacity estimation device | |
| KR20200023672A (en) | Diagnosis method of battery using Deep Learning | |
| Gasper et al. | Predicting battery capacity from impedance at varying temperature and state of charge using machine learning | |
| US20230003805A1 (en) | Method for estimating the state of an energy store | |
| CN112698216A (en) | Estimating the temperature of an electrochemical cell | |
| Park et al. | Capacity estimation of Li-ion batteries using constant current charging voltage with multilayer perceptron | |
| Kim et al. | State of health estimation of li-ion batteries using multi-input lstm with optimal sequence length | |
| CN112698214B (en) | Estimating battery state of electrochemical battery | |
| Huang et al. | An approach to measurements of electrical characteristics of lithium-ion battery with open-circuit voltage function | |
| CN112698213B (en) | Analyzing electrical impedance measurements of electrochemical cells | |
| Nenadic et al. | Estimation of state-of-charge and capacity of used lithium-ion cells | |
| CN112698212A (en) | Estimating battery state from gradients of electrical impedance measurements | |
| CN114740389A (en) | Battery health assessment method and device, electronic equipment and readable storage medium | |
| Singh et al. | State of charge estimation techniques of Li-ion battery of electric vehicles | |
| Vaidya et al. | State of temperature detection of Li-ion batteries by intelligent gray box model | |
| Shete et al. | Fault Diagnosis and State of Charge Estimation of Battery for Electric Vehicle Applications | |
| Paul et al. | Comparative Study of Different Regression Models for Estimating Lithium Ion Battery Pack Capacity | |
| CN114994553A (en) | Online identification method and system for internal resistance of battery | |
| CN117825991A (en) | Battery abnormality detection method, apparatus, device, storage medium, and program product | |
| Vaidya et al. | Surface Temperature Detection of Li-Ion Batteries by Intelligent Gray Box Model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |